Yacc: A Syntactic Analysers Generator

Yacc: A Syntactic Analysers Generator

Compiler-Construction Tools The compiler writer uses specialised tools (in addition to those normally used for software development) that produce components that can easily be integrated in the compiler and help implement various phases of a compiler. Scanner generators Parser generators Syntax-directed translation Code-generator generators Data-flow analysis: key part of code optimisation

Relationship between Parser and Scanner Parser guides scanner by asking tokens one by one. Interaction with the symbol table is crucial.

Lex and Yacc There exist various tools for generating parsers. We will discuss Yacc, the companion of Lex: the programs they generate can share variables and procedures and therefore can be compiled jointly. Lex & YACC are known to GNU/Linux users as Flex & Bison, where Flex is a Lex implementation by Vern Paxson and Bison the GNU version of YACC. Typically, the Lex actions are not visible on the standard output; rather Lex return values and control to the caller (i.e. the parser). Function yylex() returns the token names, while the token values are shared by means of global variables such as yylval of type int.

Yacc LALR(1) parser generator whose first version is due to S.C. Johnson (AT&T Bell Laboratories) in the early 1970s. To construct a translator, Yacc operates as follows: Users only need to care of the Yacc source (file.y where the grammar is defined); everything else is implemented automatically.

Yacc Basics The parser generated by yacc is a C function called yyparse( ) is an LALR(1) parser yyparse( ) calls yylex( ) repeatedly to obtain the next input token The function yylex( ) can be hand-coded in C or generated by lex yyparse( ) returns an integer value 0 is returned if parsing succeeds and end of file is reached 1 is returned if parsing fails due to a syntax error

Yacc Specifications Structure of a Yacc program: %{ C declarations %} declarations %% translation rules %% supporting C routines A production of the form is expressed in Yacc by the rules: nonterm corpo 1 corpo k nonterm : body 1 {semantic action 1} body k {semantic action k } ;

Declarations The declaration part of a Yacc program has two (both optional) sections: ordinary C declarations delimited by %{ and %} (e.g. standard header files) declarations of grammar tokens of the form %token DIGIT These declarations are made available to the analyzer generated by Lex when lex.yy.c is compiled together with the Yacc output.

Supporting C-Routines The third part of a Yacc specification contains procedures in C: The lexical analyzer is provided as yylex(). Usually this is produced by Lex. Other procedures such as error recovery routines. If the pair (token-name, attribute-value) is produced by yylex(), then token-name must be declared in the first section of the Yacc specification, while attribute-value is communicated to the parser through the variable yylval (defined by Yacc).

Attribute Values and Semantic Actions Every grammar symbol has an associated attribute value An attribute value can represent anything we choose The value of an expression The data type of an expression The translated code Yacc associates an attribute with every token and non-terminal Token attributes are returned by the scanner in the yylval variable Non-terminal attributes are computed while parsing Attribute values are pushed an popped on a semantic stack The semantic stack operates in parallel with the parser stack A semantic action in Yacc is a code fragment delimited by { } Executed when yacc matches a rule in the grammar Semantic Actions can be used to make calls to semantic routines

Example Construct a simple desk calculator that reads an aritmetic expression, evaluates it and then prints its numeric value. Start with the following grammar for arithmetic expressions: E E + T T T T F F F (E) digit The token digit is a single digit between 0 and 9.

Example ctd.

Creating Yacc analyzers The Lex library ll provides the driver program yylex() to Yacc. When using Lex and Yacc together, we can replace the routine yylex() in the third part of the Yacc specification by the statement #include "lex.yy.c" Since the Lex output file is compiled as part of the Yacc output file y.tab.c, the lexical analyzer can then have access to Yacc s names for tokens.